Identification of drug-target modules in the human protein–protein interaction network

The human protein–protein interaction network (PIN) has a modular structure, in which interactions between proteins are much denser within the same module than between different modules. Proteins within the same module tend to have closely related functions with each other. Therefore, if a module is composed of relatively small number of proteins (e.g., modules composed of less than 5 % of all proteins in the PIN) and significantly enriched with target proteins for a disease, proteins and interactions in the module are likely to play an important role in disease mechanisms and may be potential candidate targets for the disease. We defined such modules as “drug-target modules.” In order to find drug-target modules in the human PIN, we developed a novel computational approach that decomposes the network into small modules and maps drug targets on the modules. The approach successfully identified drug-target modules that contain more than 40 % of targets of cancer molecular-targeted drugs (e.g., kinase inhibitors and monoclonal antibodies). Furthermore, proteins in the modules are significantly involved in cancer-related signaling pathways (e.g., vascular endothelial growth factor signaling pathway). These results indicate that the listing of proteins and interactions in the drug-target modules may help us to search efficiently for drug action mechanisms and novel candidate targets for cancerous diseases. It may be pertinent to note here that, among proteins in the drug-target modules, proteins with a small number of interactions may be potential candidate anti-cancer targets with less severe side effects.